High prevalence and excess mortality of late presenters among HIV-1, HIV-2 and HIV-1/2 dually infected patients in Guinea-Bissau - a cohort study from West Africa

Introduction HIV infected individuals with late presentation (LP) and advanced disease (AD) have been associated with higher mortality, higher cost of medical management, impaired CD4 cell count increment and potentially ongoing risk of HIV transmission. Here we describe the proportion of patients with LP and AD at an HIV clinic in Guinea-Bissau, identify risk factors and evaluate the outcome of these patients. Methods We included all patients >15 years diagnosed with HIV-1 and/or HIV-2 at the outpatient HIV clinic at Hospital National Simão Mendes, during June 2005 - December 2013 in a retrospective cohort study. Patients were followed until December 2014. LP and AD was defined as a baseline CD4 cell count of 200-349 cells/µL and <200 cells/µL, respectively. Results A total of 3,720/5,562 (65.7%) patients had a CD4 cell count measured within the first 90 days of HIV diagnosis. Forty-eight percent had AD and 23% had LP. Risk factors for presentation with AD were male sex, age >30 years, Fula and Mandinga ethnicity. HIV-2 and HIV-1/2 dually infected patients had lower risk of AD compared with HIV-1 infected patients. Although antiretroviral therapy (ART) was initiated for 64.4% of patients, those with AD progression had a 3.82 times higher mortality compared to patients with non-LP. Conclusion The majority of HIV infected patients presented late. Most of the late-presenters had advanced disease and patients with advanced disease had a very high mortality. Initiatives to enroll patients in care at an earlier point are needed and should focus on risk groups.


Introduction
The introduction of antiretroviral treatment (ART) for HIV infection has significantly reduced the morbidity and mortality of HIV infected individuals [1]. However, the success of treatment depends on the disease progression at HIV diagnosis and adequate initiation of ART [2,3]. Low CD4 cell count at HIV diagnosis termed late presentation (LP) and late presentation with advanced disease (AD) have been associated with higher mortality [4], higher cost of medical management [5], impaired CD4 cell count increment [6,7] and potentially ongoing risk of HIV transmission [8,9]. Furthermore, lag time between CD4 cell count measurement and subsequent medical consultation where ART initiation is decided, may contribute to further disease progression and loss to follow-up [10,11]. Thus, intensified efforts are needed to identify and enroll HIV infected individuals earlier into care [12]. In West Africa, two types of HIV are circulating: HIV-1 and HIV-2. The UNAIDS estimates that in this part of the world less than half of the HIV infected individuals in need of ART are actually receiving the treatment [1]. The West African country Guinea-Bissau has the world´s highest prevalence of HIV-2 and the prevalence of HIV-1 has been rising [13]. Compared with HIV-1, HIV-2 is less transmissible, associated with a lower viral load and with a slower rate of both CD4 cell decline and clinical progression. Still, it may lead to AIDS, with clinical features indistinguishable from the syndrome caused by HIV-1 [14][15][16]. In this study, we describe the proportion of HIV-1 and/or HIV-2 infected patients with LP and AD at an HIV clinic in Bissau, to identify risk factors thereof and to evaluate the outcome of these patients.

Setting and patients
In June 2005, the first HIV clinic opened in Bissau, the capital of Guinea-Bissau, providing ART free-of-charge, and the clinic is now the largest ART center in the country in terms of patients on followup. CD4 cell count measurements have been performed since 2007, and the same year the Bissau HIV Cohort was established to evaluate patient treatment and follow-up [17]. Routines at the clinic has previously been described [18].

Definition of LP and AD
Based on a consensus definition [17] we defined LP as patients presenting for care with a CD4 cell count below 350 cells/µL. AD were defined as patients presenting with a CD4 cell count below 200 cells/µL or presenting with an AIDS-defining event, regardless of the CD4 cell count. Only patients with a CD4 cell count measured within 90 days of HIV diagnosis were included in the analyses.

Statistical methods
We compared the demographic, clinical and laboratory features of patients with AD, LP and non-LP using χ 2 test for categorical variables. Logistic regression was used for the analysis of risk factors for LP and AD patients when comparing with non-late presenters. In case of missing data, a missing data (unknown) group was made and included in the analysis to avoid exclusion of patients. Variables associated with AD or LP in the univariable model

General characteristics
A total of 5,562 patients were diagnosed with HIV in the study period, but 68 (1.2%) patients were excluded as they were already receiving ART at first CD4 cell count. Furthermore, 1,843 patients (33.1%) were excluded from the analysis as CD4 cell count was not measured within the first 90 days after HIV diagnosis. Excluded patients were more likely to be single versus being married with an odds ratio (OR) of 1.36 (p < 0.01). In addition, a higher proportion of male patients did not have a CD4 cell count measured (OR 1.23, p < 0.01). In the subsequent analysis (Table 1)

Description of CD4 cell counts
The median time between HIV diagnosis and initial CD4 cell count measurement was 1 day (IQR 1-5 days). Only a smaller proportion (18.3%) of the patients had CD4 cell count measured the same day as the HIV diagnosis, whereas 33.9% had the analysis performed the following day. Figure

Risk factors of LP and AD
Risk factors of AD in the multivariable analysis (Table 2)

Initiation of ART
ART was initiated among 2,395 (64.4%) patients during the study period, and this proportion was higher among patients with LP (76.8%) and AD (76.6%) than among patients with non-LP (33.2%), p < 0.001. The overall median time to ART initiation was 18 days (IQR: 9 -49 days). The median time to ART initiation was lowest for patients with AD (14 days) followed by patients with LP (20 days) and non-LP (266 days, p < 0.001). Only 44 (1.2%) patients initiated ART the same day as they were diagnosed with HIV.

Mortality of patients
All patients were included in the mortality analysis and contributed

Patients lost to follow-up and patients transferred
According to our definition, 2,169 (58.3%) patients became lost to follow-up during the study period. The proportion of patients lost to follow-up was higher among non-late presenters (67.8%) and patients with LP (60.5%) compared with AD (51.8%, p < 0.001). In total, 178 (4.8%) patients were transferred to another facility with no difference between groups; non-LP (4.5%), LP (5.9%) and AD cohorts are characterized by a high rate of loss to follow-up [10,19] and in a meta-analysis a combined 40% of patient lost to follow-up had in fact died. Thus loss to follow-up in our study may have underestimated the true mortality rate [20]. Another limitation is the uncertainty in the HIV typing. We have previously shown that the rapid test used in Guinea-Bissau for HIV type discrimination (SD Bioline HIV 1/2 3.0) may overestimate the number of HIV-1/2 dually infected patients [21,22] [24] and some differentiate between LP and AD whereas others do not [17]. These year [26]. In South Africa, 33.6% of 830 patients presented late (CD4 cell count <100 cells/µL), which was associated with living far away from test site, working outside the home, perceiving health service barriers and/or having poor emotional health [27]. A total of 2,311 HIV infected patients were included in a study from Uganda and 40% were late presenters (WHO disease stage 3 or 4).
Predictors of late presentation were age 46-60 years (vs. younger), lower educational level, being unemployed, living in a household with others, being unmarried and lack of spousal HIV status disclosure [28]. The numerous risk factors of LP/AD mentioned above indicates the diversity in explanatory causes of late HIV testing and demonstrates the need for local studies. Causes of LP/AD may differ with geographical region and local settings. As in our study, male sex was a risk factor of LP in many of these African studies possibly reflecting a poorer health-seeking behavior in the male population [12,[26][27][28][29]. In Bissau male sex is also associated with higher rates of loss to follow-up after CD4 cell count measurement [10].

Implications
Among all patients diagnosed with HIV at our clinic, 33.1% did not days [31]. However, the performance of different point-of-care devices for CD4 cell count varies with regard to accuracy and costs [32]. The implementation of point-of-care monitoring tools at decentralized clinic laboratories may therefore be warranted [33]. Controversially, studies have suggested that initial infection with HIV-2 may protect against subsequent HIV-1 disease progression [34,35] whereas other studies did not find a protective effect [36].
Based on the HIV tests used, we found that HIV-1/2 dually infected patients were significantly less likely to have AD than HIV-1 infected patients. A recent meta-analysis of 56 studies from Sub-Saharan Africa found that the overall median CD4 cell count in 2002 was 251 cells/µL [37]. The median CD4 count was somewhat lower in our study (206 cells/µL). Due to intermittently unavailable CD4 cell of ART in Guinea-Bissau may also explain why age >30 years was a risk factor of AD in our study period. On average, time since infection may be longer for older patients causing more severe disease progression. Low self-perceived risk of HIV infection among older patients may also explain later presentation [28]. Cultural differences between different ethnicities may influence time of presentation at an HIV clinic. Fula and Mandinga ethnicity, which were associated with AD in this study, constitutes two major population group in West Africa. In Guinea-Bissau distinct living conditions and cultural patterns are practiced in these ethnic groups, participation in vaccination campaigns are lower and child mortality is higher than in other groups [38][39][40][41]. Similarly, the mortality rate was higher in these ethnic groups among women of reproductive age [42]. HIV-1 infection was also associated with Mandinga ethnicity in a study among pregnant women in Bissau [43]. Association between AD and Fula/Mandinga ethnicity may reflect differences in geographic origin [42], socioeconomic characteristics and health seeking behavior [40]. During the study period, the WHO recommendations for initiation of ART based on implemented at the end of the study period. ART is provided freeof-charge for patients, but still, approximately 25% of patients with AD in our study did not initiate ART. Early mortality was very high for patients with AD, but may be improved with rapid initiation of ART. We have previously described that loss to follow-up bears a significant cause of the failure to start ART both before and after the first CD4 cell count [10].

Unanswered questions for future research
As patients with AD have higher mortality rates, it should be of high priority to investigate the causes of LP/AD at HIV testing. Additional

Competing interests
The authors declare no competing interest.